ELLIS UnConference 2026
European machine learning community day on 8 December 2026 at Palais des Congrès de Paris, opening the NeurIPS European satellite. €90.
The ELLIS UnConference is the annual gathering of Europe's machine learning community, organized by ELLIS. Following editions in Copenhagen, Spain, Germany, and France, the 2026 edition takes place as a single full day on 8 December 2026 at Palais des Congrès de Paris, kicking off the Paris satellite of NeurIPS.
Its "unconference" format puts the community in charge: rather than a top-down speaker lineup, the schedule is built directly from sessions proposed by researchers and practitioners.
Organizers
- ELLIS — European Laboratory for Learning and Intelligent Systems, a pan-European network of AI labs and researchers
- Organizing Committee: Maxime Haddouche, Marie Generali Lince, Léo Dana, Matthieu Blanke, Sacha Braun, Eugène Berta, Ambroise Odonnat, Félix Chavelli
- Advisory Committee: Edwige Cyffers, Benjamin Guedj, Søren Hauberg, Conrad Philipp
Key numbers
- 20 two-hour sessions, ten in the morning and ten in the afternoon
- 1 opening keynote
- 1 poster session closing the day
Program
| Time | Session |
|---|---|
| 08:00–08:30 | Welcome & badge pickup |
| 08:30–09:00 | Opening remarks |
| 09:00–10:00 | Keynote by Carina Prunkl |
| 10:30–12:30 | Morning sessions (10 in parallel) |
| 14:00–16:00 | Afternoon sessions (10 in parallel) |
| 16:30–18:30 | Poster session |
Keynote
- Carina Prunkl — "The Evaluation Gap and the Governance of General-Purpose AI". Drawing on the International AI Safety Report 2026, the talk examines the widening gap between what current model benchmarks measure and what decision-makers need to assess real-world behavior.
Morning sessions (10:30 am – 12:30 pm)
- "What Should a World Model Remember?" — Randall Balestriero, Julia Kempe, Yann LeCun (World Models)
- "Tabular Foundation Models: Principles, Practice, and Open Problems" — Frank Hutter, Marine Le Morvan
- "Multimodality for Everyone (M4E): Democratizing Multimodal Foundation Models" — Sébastien Bratières, Yiannis Kompatsiaris, Luisa Bentivogli et al.
- "Health Foundation Models: From Prediction to Decision-Making and Self-Improvement" — Jiancheng Yang, Kimmo Kartasalo, Azade Farshad, Koen Van Leemput
- "Are We at the Tipping Point for Agentic AI Deployment?" — Nuria Oliver, Novi Quadrianto, Adrian Weller, Plamen P. Angelov, Javier Sanguino
- "Long-Horizon AI Agents: Reasoning, Recovery, and Coordination" — Beyza Ermis, Sharon Li
- "Why Causality and Uncertainty Must Converge in Practice" — Irene Balelli, Anna Calissano, Marianne Clausel, Emilie Devijver, Alex Markham, Adele Ribeiro
- "Open-Source Foundation Model Training in Europe" — Aaron Klein, Michael Tiemann, Frank Hutter, David Salinas et al.
- "Multi-Agent AI for Scientific Discovery: Learning, Coordination and Evaluation" — Yali Du, Jun Wang
- "Constrained Learning for Scientific Modelling" — Cecília Coelho, Oliver Niggemann et al.
Afternoon sessions (2:00 pm – 4:00 pm)
- "Simulated vs. Unsimulated World Models" — Sang Wan Lee, Mattia Rigotti et al.
- "AI for Tabular & Relational Data: What's Next?" — Lennart Purucker, Madelon Hulsebos, Félix Lefebvre et al.
- "Rethinking AI: Sustainability and Frugality in the Era of Foundation Models" — Quentin Bouniot, Alexandre Allauzen, Edouard Oyallon, Florence d'Alché-Buc et al.
- "AI for Earth and Climate Sciences" — Gustau Camps-Valls, Markus Reichstein et al.
- "Reflections on XAI" — Nicolò Cesa-Bianchi, Ulrike von Luxburg et al.
- "UniReps: Unifying Representations in Neural Models" — Samuel Lippl, Clémentine Dominé et al.
- "Amortized Probabilistic ML" — Herilalaina Rakotoarison, Julia Linhart, Katharina Eggensperger, Luigi Acerbi
- "ELLIS Innovation Area Kick-off: From Frontier Research to Impact" — Matthias Bethge, Karteek Alahari, Nuria Oliver, Fabian Theis et al.
- "Next Generation of Discovery Loops" — Samuel Kaski, Chris Holmes, Yee Whye Teh et al.
- "Learning on Graphs for Scientific Discovery" — Daniel Daza, Victorita Dolean, Mykola Pechenizkiy
The complete speaker list for every session is published on the UnConference website.
Poster session
Attendees may present a paper accepted in 2026 at a top conference or journal (ICML, ICLR, AISTATS, JMLR, etc.), excluding NeurIPS 2026 papers. Submission deadline: 14 November 2026; notification: 20 November 2026. A1 portrait format, printed by participants.
Pricing
| Profile | Price | Included |
|---|---|---|
| Student / Academic / Industry | €90 incl. VAT | Full-day access, coffee breaks |
Lunch is not provided.
Why attend?
- An open-format European ML research day built around two-hour interactive discussions rather than one-way lectures.
- Multiple sessions directly intersect data and AI leadership priorities: agentic AI deployment, European open-source foundation model training, model frugality, and explainability.
- A flat €90 ticket — a fraction of typical conference fees — serving as the gateway to NeurIPS week in Paris.
Partners
Gold Sponsors:
- G-Research
- Prior Labs
- MBZUAI
Silver Sponsor:
- QRT (Qube Research & Technologies)
Practical information
- Date: Tuesday 8 December 2026, 8:00 am – 6:30 pm
- Venue: Palais des Congrès de Paris, 2 place de la Porte Maillot, 75017 Paris
- Access: Metro Line 1, Porte Maillot station
- Pricing: €90 incl. VAT
- Registration: ellisunconference2026.dakini-pco.com
- Official site: ellis.eu
Frequently Asked Questions
When does ELLIS UnConference 2026 take place?
Where is ELLIS UnConference 2026 held?
How much is registration?
How does an 'unconference' work?
Can participants present a poster?
Who organizes ELLIS UnConference?
Let's work together
A digital transformation project, a marketplace to launch or an AI strategy to frame? Explore my background or reach out directly.


